Dan Nicholson <[email protected]> writes: > On Mon, Nov 9, 2009 at 9:52 AM, tom fogal <[email protected]> wrote: > > Mark Kettenis <[email protected]> writes: > >> > From: Adam Jackson <[email protected]> > >> > Date: Mon, 09 Nov 2009 10:51:13 -0500 > >> > > >> > On Sat, 2009-11-07 at 16:29 -0800, Ian Romanick wrote: > >> > > >> > > I vaguely recall there being some interaction with the X > >> > > server. If Mesa is built for TLS, then the X server must also > >> > > be built for TLS. [. . .] [snip] > > In the case that we're building against an X server, is there > > perhaps some kind of configure-time test we could do? Â Then the > > Mesa default could be `autodetect' instead. > > Make it a variable in gl.pc and query it from the xserver configure.
Sorry, that was ambiguous; I wasn't trying to ask how this might be done. I was wondering if the X server already provided this information in some way Mesa could figure it out. I imagine someone would have piped up if such support already existed though, so I guess I should volunteer. > gl.pc: > ... > glx_tls = yes > > xserver/configure: > if test "x$GLX_USE_TLS" = xauto; then > GLX_USE_TLS=`$PKG_CONFIG --variable=glx_tls gl` > test -z "$GLX_USE_TLS" && GLX_USE_TLS=no > fi .. but now that you mention it, wouldn't it be the reverse? That is, we'd want the X server to provide a .pc with `tls = something', and have Mesa query the server's .pc? I guess it depends on which is `lower' in the software stack, which I've been assuming is the X server (correct me?). -tom
_______________________________________________ xorg-devel mailing list [email protected] http://lists.x.org/mailman/listinfo/xorg-devel
